DataBalanceAnalysis

DataBalanceAnalysis contains a set of metrics for diagnosing and measuring data imbalance. This module is intended to be used as part of the error diagnosis process for failure modes that are due to class or feature imbalance. After measuring with DataBalanceAnalysis, AI practitioners can then work to mitigate the failure through techniques available in the library’s DataProcessing module.

Example

A model trained for house-price prediction is discovered to be underperforming for houses that do not have an attached garage. The AI practitioner determines that this failure is due to the underrepresentation of houses with no garage in the training data. The practitioner can use metrics in the DataBalanceAnalysis module to measure the feature imbalance (“garage” vs. “no garage”), then work to mitigate the issue by using one of the sampling techniques available in the library’s DataProcessing module for augmenting data.

API

• DataBalanceAnalysis
  • AggregateBalanceMeasure
  • DistributionBalanceMeasure
  • FeatureBalanceMeasure

Examples

All notebooks listed here can be found in the folder notebooks/.

Data Balance Analysis using the Adult Census Income dataset

Python Data Balance Analysis

End to End Data Balance and Error Mitigation